Costs and benefits of controlling Campylobacter in the Netherlands - integrating risk analysis, epidemiology and economics.

(1)

1_{National Institute for Public Health and the Environment, P.O. Box 1, 3720 BA Bilthoven,} 2_{Wageningen UR, Agricultural Economics Research Institute, P.O. Box 29703, 2502 LS Den Haag,}

3_{Wageningen UR, Animal Sciences Group, P.O. Box 65, 8200 AB Lelystad,} 4_{Wageningen UR, RIKILT, P.O. Box 230, 6700 AE Wageningen}

Food and Consumer Product Safety Authority, P.O. Box 202, 7200 AE Zutphen the Netherlands

*_{Contact: A.H. Havelaar}

Microbiological Laboratory for Health Protection e-mail arie.havelaar@rivm.nl

RIVM report 250911009/2005

Costs and benefits of controlling Campylobacter in the Netherlands Integrating risk analysis, epidemiology and economics

A.H. Havelaar1*, M.J. Nauta1, M.-J.J. Mangen1, 2, A.G. de Koeijer3, M.-J. Bogaardt2, E.G. Evers1, W.F. Jacobs-Reitsma2, 4, W. van Pelt1,

J.A. Wagenaar3 , G.A. de Wit1, H. van der Zee5

This investigation has been performed by order and for the account of the Ministry of Public Health, Welfare and Sports and the Ministry of Agriculture, Nature and Food Quality, within the framework of project 250911, CARMA, Campylobacter Risk Management and

Assessment.

(2)

Abstract

Costs and benefits of controlling Campylobacter in the Netherlands – integrating risk analysis, epidemiology and economics.

A combination of decontamination with a chemical such as lactic acid and technical measures to reduce leakage of feces during slaughtering have been shown by model calculations in a Netherlands study to be the most economic method to improve the safety of broiler meat. Campylobacter bacteria form the most common bacterial cause of foodborne infections in the Netherlands, with approximately 80,000 cases of gastroenteritis per year. Of the many

different routes by which humans can be exposed to Campylobacter, the most important are the consumption of broiler meat and other raw food products, and direct contact with animals. Results above are derived from a multidisciplinary study on the costs and benefits of

measures to reduce the contamination of broiler meat. Model calculations also showed

additional hygienic measures to theoretically reduce the contamination on broiler farms, but it is not yet clear what exact measures should be taken. In the short term, more effect can be expected from additional measures in the processing plant to reduce the level of

contamination of meat. Model calculations indicate that in the Netherlands alone, this would result in the prevention of 12,000 cases of gastrointestinal illness per year. Successful

implementation of these measures will require additional studies on a practical scale. Because of unfamiliarity and additional costs, there is little support for measures among consumers and industry. Therefore active communication will be paramount.

Keywords: Campylobacter, exposure routes, prevention, broiler chicken meat, cost-utility ratio, societal support

(3)

Het rapport in het kort

Kosten en baten van Campylobacter bestrijding in Nederland – integratie van risico-analyse, epidemiologie en economie

Een combinatie van decontaminatie met een middel als melkzuur en technische maatregelen om de verspreiding van mest tijdens het slachten tegen te gaan, lijkt volgens

modelberekeningen de meest economische methode om de microbiologische veiligheid van kippenvlees te verbeteren. Campylobacter-bacteriën zijn de belangrijkste bacteriële

veroorzakers van voedselinfecties in Nederland, met ongeveer 80.000 gevallen van gastro-enteritis per jaar. Onder de vele verschillende routes waarlangs de mens aan Campylobacter kan worden blootgesteld nemen consumptie van kippenvlees, direct contact met dieren en rauw geconsumeerde producten een belangrijke plaats in. Genoemde resultaten zijn

verkregen in een multidisciplinair onderzoek naar de kosten en baten van maatregelen om de besmetting van kippenvlees terug te dringen. Aanvullende hygiënemaatregelen op de

boerderij zouden volgens modelberekeningen in theorie de besmetting bij het pluimvee sterk terug kunnen brengen, maar het is nog niet duidelijk welke maatregelen precies genomen moeten worden. Op korte termijn is meer effect te verwachten van aanvullende maatregelen op het slachthuis om de besmettingsgraad van het vlees te verminderen. Volgens

modelberekeningen kunnen alleen al daardoor in Nederland ongeveer 12.000 gevallen per jaar van gastro-enteritis worden voorkomen. Om deze maatregelen succesvol te kunnen invoeren is nog wel aanvullend praktijkonderzoek nodig. Door onbekendheid met de maatregelen en de additionele kosten is het draagvlak bij de consument en ketenpartijen gering zodat actieve communicatie van groot belang is.

Trefwoorden: Campylobacter, blootstellingsroutes, preventie, kippenvlees, kosten-utiliteit, maatschappelijk draagvlak

(4)

Preface

The CARMA (Campylobacter Risk Management and Assessment) project, which has been performed throughout the years 2001-2004, is a collaboration between the National Institute for Public Health and the Environment (RIVM), the Animal Sciences Group-Lelystad (ASG Lelystad), the Agricultural Economics Research Institute (LEI), the Food and Consumer Product Safety Authority (VWA/KvW) and RIKILT – Institute of Food Safety. The full results of the project have been documented in a number of reports and publications.

Appendix 1 contains an overview of these documents, which are also accessible through the website www.rivm.nl/carma. These reports are not further referred to in this text. There are referrals to other sources in the text. The reports and publications contain an overview of all the collaborators who have worked on the project and provide detailed background

information.

This report is a summary of the most important results of the project, written for a broader audience than the underlying technical reports. It is a translation of a report originally written in Dutch (RIVM report 250911008). The report builds on the contribution of all project team members. Besides the authors of the report a contribution to the underlying technical reports was made by Elly Katsma, Egil Fisher, Mart de Jong en Fimme-Jan van der Wal (ASG Lelystad), Krijn Poppe, Peter van Horne, Hans-Peter Folbert, Sandra van der Kroon, Marjolijn Smit and Hubert Sengers (LEI, The Hague), Roger Cooke and Louis Goossens (Delft University of Technology), Rinske van Koningsveld (Erasmus Medical Centre, Rotterdam), Rob Bernsen (Jeroen Bosch Hospital, ′s Hertogenbosch), Trudy Wassenaar (Molecular Microbiology and Genomics Consultants, Zotzenheim, Germany), Hans van de Kerkhof (GGD Zuid Holland Zuid, Dordrecht), Wilfrid van Pelt, Sido Mylius, Ine van der Fels-Klerx, Peter Teunis, Ardine de Wit and Winette van den Brandhof (National Institute for Public Health and the Environment, Bilthoven). Harry Verkleij (RIVM) contributed to the editing of this summary report. The authors are grateful to Noël Peters (RIVM) who translated the original Dutch report into English.

Ton van Gaasbeek (LEI) was strongly involved with the organization of this project. Regrettably, Ton has passed away shortly afterwards.

Appendix 2 gives an overview of the members of the Project Steering Committee and the Industry Forum. Besides this many experts from the field were consulted. Their contribution is accounted for in the technical reports.

In Appendix 3 definitions are given of a number of concepts used in this report and the abbreviations are further explained.

(5)

Summary

Campylobacter-bacteria are an important cause of food borne infections. There are approximately 80,000 cases of gastro-enteritis in the Netherlands per year, but also more serious syndromes and about 30 deaths. There are many different exposure routes. One important exposure route is consumption of chicken meat (20-40% of all cases of illness or 16,000-32,000 cases of gastro-enteritis). Other important exposure routes are direct contact with animals and food products that are consumed raw. In the CARMA project an estimate has been made of the possible costs and benefits of a large number of possible interventions to decrease human exposure to Campylobacter by consumption of chicken meat. The costs of intervention have been calculated in euro, the benefits in the decrease of disability adjusted life years (DALYs) and of cost of illness in euro. The focus was on interventions at broiler farms and in the processing plant. Attention was also paid to consumer information. Epidemiological, microbiological and economic knowledge as well as mathematical risk assessment models were used for the calculations.

A few potential interventions are interesting, considering their (theoretical) effectiveness and efficiency. Not a single evaluated intervention can be introduced directly or without any further conditions. There is intensive import and export of both live animals and chicken meat. It is not clear what the exact origin of chicken meat consumed in the Netherlands is. Both from a public health perspective and for the competitiveness of the Dutch broiler industry, it is advisable to implement measures at the European level, while taking import from third countries into account.

Reduction of contamination at broiler farms could be efficient in theory. However it is unclear which hygienic measures need to be taken, because the exposure routes on the farm are still unclear. The costs of hygienic measures can be very high. It is recommended that the hygiene at the broiler farm is gradually improved, starting with the implementation on all farms of existing hygiene standards and promoting their consistent use. The treatment of infected broiler flocks with bacteriophages has experimentally proven to be effective and could also be cost-efficient, if the effectiveness is confirmed in practice.

Since a major decrease of infections at the broiler farm is not expected in the short term, additional measures in the processing plant will also prove necessary. A complete ban on selling meat (as fresh product) from contaminated flocks would lead to a shortage of fresh meat in the summer period and large economic losses for industry. At this moment,

guaranteed Campylobacter-free chicken meat at the retail level is not achievable. Irradiation of all the produced meat would be too expensive. The most promising interventions in the processing plant are: limiting fecal leakage during scalding and defeathering, and the separation of contaminated and non-contaminated flocks (scheduling), followed by

decontamination of the contaminated flock using a product such as lactic acid. New (faster and more sensitive) test methods to detect Campylobacter infection in broilers flocks are a prerequisite for successful scheduling scenarios. Methods are being developed, but a validated test is not directly available. In addition there are questions concerning the

efficiency of decontamination methods in practice and about the effect on the appearance of the products which may make them (more) difficult to sell. Other methods to decrease the contamination of meat of infected flocks such as crust-freezing and heat treatment are more expensive and/or less effective than decontamination.

There is little support for additional measures to control Campylobacter in the Dutch society. Chain partners hold each other responsible. Poultry farmers and processing plants are not convinced of the favorable effects of additional measures and fear increased costs and sales problems. A majority of consumers does not worry about contamination of chicken meat with Campylobacter. They have not yet defined their position with respect to the proposed

(7)

measures due to a lack of information. There seems to be growing acceptance of irradiation. Good communication with all stakeholders is paramount during policy making as well as the introduction of measures.

(8)

1. Introduction

Reducing food borne infections and –intoxications is one of the spearheads of national and international public health policy. The attention is mainly focused on Salmonella and Campylobacter in food of animal origin. In the Netherlands, despite political attention and efforts from the industry, the contamination of poultry meat with Campylobacter remained at a high level at the start of the project in 2001 and also the incidence of human illness had not substantially change in the preceding five years. This in contrast to Salmonella, where there is a decreasing trend of

contamination of poultry meat. Also the incidence of human salmonellosis (as a consequence of contaminated poultry meat and other sources) is continuously decreasing [17].

• In 1995 the former State Secretary of the Ministry of Public Health, Welfare and Sports (VWS) announced the intention to take measures that in 5 years would lead to a reduction of 50% of the incidence of human gastro-enteritis as a consequence of contaminated food from animal origin.

• This policy plan has led in 1997 to a “Plan of Approach Salmonella and

Campylobacter in the Poultry Industry” formulated by the Commodity Board for Meat, Poultry and Eggs (PVE). Trade and industry accepted an obligation to reduce the level of contamination of Salmonella and Campylobacter in chicken meat directly after processing/cutting to less than 15% on 1 November 1999 at the latest.

• In 2000 it was established that this goal was not achieved. Also on the account of the advice “Food borne infections” of the Health Council, the Ministry of Public Health, Welfare and Sports (VWS) as well as the Ministry of Agriculture, Nature and Food Quality (LNV) have announced further measures.

• On 1 August 2001 a temporary measure was registered in the “Warenwetbesluit Bereiding en Behandeling van Levensmiddelen”. This measure obligates trade and industry to accommodate fresh poultry meat which is delivered to consumers with a label that warns against pathogenic bacteria.

• In 2001 the PVE published an “Action plan Salmonella and Campylobacter 2000+” in which the sector states how the agreed goals can still be achieved. In contradiction to Salmonella no goals were published for Campylobacter. The main reason for this was an insufficient apprehension of the possibilities of reducing Campylobacter

contamination in the primary sector. For this reason further investigations were recommended, also in the possibilities of controlling contamination further in the chain. These research questions were also the motivation for the CARMA

(Campylobacter Risk Management and Assessment) project.

• At this moment the Dutch government is considering a prohibition of Salmonella and Campylobacter on fresh poultry meat upon delivery to the consumer. This prohibition which should be of effect in 2007 should concern both imported and domestically produced meat [1]. Because the presence of bacteria is not totally avoidable on fresh meat, a “presence on a low level (zero+)” is considered. Notification of this intention to the European Committee will be made before the end of the year 2006.

(9)

The goal of the CARMA project is to advise the Dutch government about the

effectiveness and efficiency of measures aimed at reducing campylobacteriosis in the Dutch population. To this aim, two key questions are being investigated:

• Which are the most important routes by which the Dutch population is exposed to Campylobacter, and can the contribution of these routes be quantified?

• Which measures can be taken to reduce the exposure to Campylobacter, what is their expected effectiveness, efficiency, and societal support?

Campylobacter infections lead to a substantial loss of health and costs of illness among the Dutch population each year.

• It is estimated that in the Netherlands (population 16 million), there are 350,000-750,000 cases of foodborne infections by known pathogens each year. Four pathogens are responsible for approximately half of these cases: Campylobacter spp., Salmonella spp., Clostridium perfringens and noroviruses [17].

• Based on a large-scale population-based study in 1999 (Sensor [7]) it is estimated that each year 80,000 cases of Campylobacter infections result in 18,000 patients visiting there family doctor [6]. It is estimated that more than 600 hospitalizations occur and 30, mainly elderly, patients die.

• Besides gastro-enteritis Campylobacter infections often lead to more serious syndromes. It is estimated that there are 60 cases of the Guillain-Barré syndrome (GBS, a serious paralysis), that are caused by the immune response against

Campylobacter each year. This immune response also induces about 1,400 cases of arthritis (reactive arthritis). There are indications that Campylobacter infections can lead to inflammatory bowel disease (IBD) with an estimated 10 patients per year. • Most cases of illness are reported in the summer period and among children less than

5 years of age just like among young adults (18-29 years). There is no difference in incidence of the illness between men and women. The incidence is lower in large cities than in the country side and in medium-sized cities.

• Based on laboratory surveillance it is concluded that within the past eight years a decrease of incidence of campylobacteriosis has occurred with about 30% [18]. Also in surrounding countries such as the United Kingdom and Denmark, for a number of years a decrease of incidence of campylobacteriosis has been noticed. In some

countries (Belgium, Spain, France) [5] an increase has been reported, but that can also be caused by better diagnostics. There are fluctuations per year, such as in 2003 when there were substantially less cases of illness than expected based on the average trend. This could be related to the avian influenza crisis whereby consumption of chicken meat was temporarily lower [19].

• The disease burden from Campylobacter infections is estimated at about

1,200 Disability Adjusted Life Years (DALYs) per year, and is comparable with that of tuberculosis and bacterial meningitis, see Table 1. The most important

contributions to the disease burden are deaths caused by gastro-enteritis and loss of quality of life as a result of gastro-enteritis and the residual symptoms of GBS. • Costs of illness because of campylobacteriosis are estimated at about 21 million euro

(10)

Table 1. Disease burden of campylobacteriosis in comparison with other (infectious) diseases.

Amount of lost DALYs per year

Infectious diseases Non-infectious diseases

> 100,000 Cardiovascular disease,

Cancer, Depression, Diabetes, Alcohol dependancy 30,000-100,000 Pneumonia and acute

bronchi(oli)tis Traffic accidents Breast cancer Suicide 10,000-30,000 Influenza Epilepsy Multiple sclerosis 3,000-10,000 HIV/AIDS

Upper respiratory tract infections Stomach ulcers

Ulcers of the stomach and duodenum 1,000-3,000 CAMPYLOBACTERIOSIS Bacterial meningitis Bacterial SOA Tuberculosis Hip fractures < 1,000 Shiga-toxin producing

Escherichia coli O157 Source: RIVM

(11)

2. Exposure routes

The reservoirs of Campylobacter can be found in the animal world,( farm and domestic animals as well as animals living in the wild) and in the environment (water). There are many different routes by which humans can be exposed to

Campylobacter. Which routes are most important in the Netherlands, was investigated with the help of epidemiological studies and exposure assessment. The most important routes are food (particularly chicken meat and food products that are consumed raw) and direct contact with animals. Also foreign travel is a risk factor. The exact

contribution of different exposure routes is difficult to quantify. In the Netherlands drinking water plays a minor role, in contrast to what is found in some international studies.

• Two methods were used to estimate the fraction of cases attributable to consumption of chicken meat. One method was based on epidemiology (a case-control study), the second method was based on risk modeling (comparative exposure assessment). The results of these two methods were partly in agreement but there were also differences. • In a large-scale case-control study (CaSa [8]) the following factors were associated

with a higher risk of campylobacteriosis: consumption of chicken meat, not well-done meat and meat prepared on the barbecue or grill or microwave oven, eating food in a restaurant, having young dogs and cats, consumption of raw shellfish, professional contact with raw meat and contact with persons with symptoms of gastro-enteritis. People who used antacids were more often ill than others.

• Also foreign travel was associated with an increased risk, but this finding might be biased by the way patients are selected. Doctors have a tendency to submit a fecal sample for diagnosis more quickly if a patient has recently been abroad.

• Diverse protective factors were also found, although the interpretation of these factors is questionable. Consumption of various foods (meat pastries (for example croquette, sausage roll), fish, hardboiled eggs, dairy products (other than milk and cheese), salad, fruits with peel, chocolate and nuts); contact with feces of animals and visiting other households with pets was associated with a reduced probability of

campylobacteriosis. To what extent these factors are really protective is unclear. They can also be related to other habits and lifestyle factors, such as little dietary variation. • The exposure of the Dutch population via different routes was calculated based on the

occurrence of Campylobacter in diverse (food) sources and on the degree to which the population gets in contact with contaminated sources. Because of the fact that much necessary information is missing, or very limited, there is major uncertainty in these calculations. Nevertheless it can be concluded that particularly direct contact with infected animals and consumption of unheated food (such as vegetables and fruits, raw fish, and non-pasteurized milk) are important exposure routes.

The calculated incidence of cases of illness caused by Campylobacter based on exposure- and dose-response models is much higher than the incidence measured in epidemiological studies. These differences can not yet be explained. The

epidemiological information is assumed to be more reliable and is in this study used as the basis of the calculations of the disease burden and costs of illness.

• Although there are uncertainties in both the risk assessment and the epidemiological assessments, the detected differences are substantial enough that they can be

considered of essential importance and require further study.

• Different explanations are possible for the overestimation of the risk models: differences between Campylobacter strains, differences in exposed populations,

(12)

effects of the matrix in which the bacteria occur, clustering of the exposure or partial protection of the exposed population by immunity.

• The dose-response relation may overestimate the risk of infection because there is a large difference between Campylobacter-strains regarding infectiveness and

pathogenicity. It has been assumed that there is one dose-response relation that is valid for all Campylobacter strains. The information used for the dose-response model may represent the effect of highly virulent Campylobacter strains. Despite the

availability of a large number of typing methods it is still not possible to group

Campylobacter strains based on their virulence. This is caused by the fact that it is not yet clear how Campylobacter causes illness, and also because of the instability of the genome of Campylobacter, with the result of a large genetic diversity.

• Exposure and infection can lead to an immune response, which can lead to a temporary protection against re-infection and/or disease. Preliminary modeling has shown that the calculated incidence becomes much lower when such protection is taken into account. It has also been shown that the results of case-control studies are strongly disturbed as a result of immunity. The effect of immunity predominantly play a part among populations who are relatively frequently exposed to Campylobacter, for example through professional contact. But also for the general population immunity would result in a certain degree of protection. On the other hand the differences between Campylobacter strains can be a reason that immunological protection is only partial.

• A better understanding of the dynamic interaction between bacteria and their hosts is necessary for better risk assessments and to better interpret the results of

epidemiological studies.

Contaminated chicken meat is an important source of human exposure in the

Netherlands. It is estimated that at least 20% (with a maximum of 40%) of all cases of campylobacteriosis is directly or indirectly caused by contaminated chicken meat.

• In the CaSa case-control study 16% of the cases of illness are attributed to recent foreign travel.

• Of the indigenous infections (84% of the total) 23% was attributed to consumption of chicken meat. This corresponds to 19% of all cases of campylobacteriosis. This estimation is probably too low, because the contribution of foreign travel is overestimated.

• Hence, of all cases of illness caused by Campylobacter in the Netherlands at least 20% is attributable to consumption of contaminated chicken meat. So at least 16,000 cases of gastro-enteritis per year, 12 cases of GBS, 280 cases of reactive arthritis and 2 cases of IBD. The corresponding disease burden is at least 240 DALYs per year and the costs of illness are at least 4.2 million euro per year.

• The upper limit of all cases of campylobacteriosis caused by contaminated chicken meat is derived from a Belgian study [20]. During the dioxin crisis of 1999 sales of chicken meat were prohibited during a period of four weeks. In this period the incidence of campylobacteriosis was 40% lower than expected based on previous years, to then return to the normal level. This percentage is considered as the upper limit of the fraction caused by chicken meat in the Netherlands. The incidence of illness, disease burden and costs of illness can therefore be twice as high as mentioned above.

(13)

3. Research methods

In consultation with representatives of the Ministries of VWS and LNV, of the VWA and of the industry a number of interventions in the chicken meat chain were studied that could possibly lead to a reduction of the health risk of the consumer. With the help of mathematical and economical models the possible costs and benefits were estimated. Attention was also paid to the societal support for these measures among consumers and the industry. This information aims to support the political decision – making process (see Figure 1).

Epidemiological studies -human -animal

Risk assessment _{Intervention scenarios}

Disease burden Cost of illness Cost - utility analysis Acceptability measures Data - observational -experimental number of infections Autonomous developments measures

Supply chain costs

Risk management decision making process Political culture/ limitations

(14)

The mathematical model for risk assessments consists of several modules: farm, processing plant, consumption and disease.

• The chosen interventions are directed against contamination with Campylobacter in different stages of the chicken meat chain, see Figure 2. These correspond with the modules in the mathematical model.

• The farm module describes the rate at which an infection spreads in a broiler house, the relationship between the probability of infection of the animals with

Campylobacter and the number of houses at a farm, and in how far there is a relationship between infections of chicks in consecutive production cycles.

• The processing plant module describes the changes of the number of campylobacters on a carcass of chickens as a result of spreading of contamination through the

processing environment. The model differentiates the following phases: scalding, defeathering, evisceration, washing, cooling and cutting. The calculations focus on the contamination of fillet of chicken, because this product is consumed often in the Netherlands and handled in the kitchen in such a way that cross-contamination to salads and the like can take place.

• The distribution module describes the die-off of Campylobacter during transport and storage in retail.

• The preparation module concerns exclusively the cross-contamination between a chicken breast and a salad consumed at the same meal. This route is considered representative for a large part of the chicken meat related contaminations.

• The consumption module describes the exposure of the consumer to Campylobacter by cross-contamination of salads taking consumption information into account. It is assumed that exposure from undercooked chicken meat is less important than exposure by cross-contamination.

• The disease module predicts the number of cases of illness as a result of consuming cross-contaminated salads based on a dose-response relation.

• The model accounts first for the situation in the Netherlands in the year 2000. This year was chosen at the start of the project as base year for the calculations because the information of that year is considered representative for the situation in the

Netherlands. In 2000 there was no interference in the market caused by food crises or outbreaks of contagious animal diseases.

• Subsequently, the model is used to estimate the expected results of each of the chosen interventions, by changing the values of relevant parameters in the model.

• The relation between the numbers of predicted cases of illness in an intervention scenario compared to that of the baseline scenario is a measure for the effectiveness of the intervention. This refers to consumers of chicken meat produced from broilers held in the Netherlands. It is assumed that effects calculated for the contamination route chicken breast fillet → salad can be extrapolated to all cases of illness associated with consumption of chicken meat.

• Based on the predicted reduction of disease-incidence also the reduction of disease burden (averted DALYs) and costs of illness are being calculated. These values are later compared to the costs of the implementation of an intervention. To calculate the costs of an intervention investments were, as usual in economic studies, discounted using a rate of 4%. Because of comparability also the disease burden and costs of illness were discounted at the same rate, leading to respectively 170 DALYs and 3.9 million euro per year as a result of consumption of chicken meat. Discounting the disease burden reflects that people in general value their current health state higher than a similar health state in future because there is uncertainty about future

(15)

• The cost-utility of an intervention is being calculated as the ratio between net costs of an intervention (i.e. the costs of implementation reduced with the averted costs of illness) and the averted disease burden in DALYs.

• The results of these calculations have been summarized in Table 3 (see page 38) and are explained in the following text. For many calculations information was lacking, so they had to be based on estimates of experts and of the researchers. Because of this, sensitivity analyses have been performed. For the most important uncertainties, besides the most likely value also an optimistic and pessimistic estimate is given and the effects are being analyzed. The results of the calculations in the text are generally the result of the most likely estimates. Besides that the most important results of the sensitivity analyses are being explained.

Farm

Processing

Consumer

Logistic slaughter Scheduled treatment Reduction of faecal leakage Decontamination of scalding tank

Treatment of carcasses:

decontamination, crust freezing, irradiation Freezing of products

Channeling to meat processing

Freezing at home Improved kitchen hygiene

Monospecies farms Improved hygiene No thinning Phage therapy Information campaign

(16)

4. The significance of import and export

Only a part of the consumed chicken meat in the Netherlands originates from

domestically held and processed broilers. This means that measures on the farm or in the processing plant have less effect on Dutch public health if they are not also taken in exporting countries. A large part of the chicken meat produced in the Netherlands is being exported. Measures in the Netherlands may also lead to fewer cases of illness in other countries. It turns out to be not possible to get a quantitative insight in the trade volumes concerned, especially to what extent imported meat is also exported again. This leads to uncertainty about the consequences of measures. Also the comparison of costs and benefits of measures becomes more difficult.

• According to information from the Commodity Board for Poultry and Eggs (PVE) in the base year (2000) 617,000 tons of broilers (converted to processing weight) were produced. Of that 9,000 tons was exported as living animals but also 68,000 tons of living animals were imported. So net 676,000 tons was processed, of which 90% from animals held in the Netherlands, (See Figure 3).

• In 2000 577,000 tons of chicken meat was exported, but also 157 tons was imported. So in the Netherlands 256,000 tons of chicken meat was consumed. It is unclear which part of the imported chicken meat is also exported again (the transit factor), and therefore does not end on the Dutch market. There is also crossing-border transport within one company, e.g. broilers that were held just over the Dutch border are processed and sold in the Netherlands.

• If we assume that all fresh imported meat (40%) also ends up on the Dutch market, than 68% of the in the Netherlands consumed meat originates from domestically held and processed broilers. 7% of consumed meat originates from animals held abroad but processed in the Netherlands. Measures on Dutch farms have no effect on the

contamination of this meat, but measures in Dutch processing plants do have an effect. 25% of consumed meat originates from animals held and processed outside of the Netherlands. Measures in Dutch processing plants have no effect on the

contamination of this meat. Measures that are focused on the consumer have an effect on all meat, regardless of origin.

• The maximum achievable effect of interventions on Dutch farms is 68% of all 16,000 cases of gastro-enteritis caused by consumption of chicken meat, i.e. about 11,000 cases.

• With measures in Dutch processing plants a maximum of 75% or 12,000 cases of gastro-enteritis can be prevented. The remaining cases can only be prevented if measures adopted in the Netherlands also apply for the producers of imported meat or by measures focused on the consumer.

• Measures on Dutch farms and in processing plants are not only favorable for the Dutch consumer but also for foreign consumers of meat produced in the Netherlands. If all consumers of chicken meat produced in the Netherlands are considered, than the cost-utility of the measures taken in the Netherlands is more favorable.

(17)

Gross domestic production 617 Net slaughter weight 676 Domestic consumption 256 Imported live birds

Imported meat and meat

products 157

Exported live birds

Exported meat and meat products 577 94 63 483 FARM PROCESSING CONSUMPTION 68 9 193

Figure 3. Import and export of broiler- and chicken meat in the Netherlands in the year 2000. Source: PVE. All data in 1000 tons (live) processed weight. It is assumed that 40% of the imported chicken meat (the fresh part) ends up on the Dutch market.

(18)

5. Interventions on the farm and during transport

Broilers are contaminated by Campylobacter on the farm. It is still unclear which factors precisely play a role. It is clear however that it takes strict hygiene measures to keep the risk of contamination as small as possible, but that strict even application of known hygiene measures cannot guarantee the production of Campylobacter-free broiler flocks.

• According to information of the PVE in the Netherlands, in the base year 2000, upon delivery at the processing plant, an average of 36% of the processed flocks of broilers was infected with Campylobacter. Among these flocks the prevalence of

contaminated animals is nearly always larger than 90%. There is a number of flocks in which contamination is introduced shortly before delivery at the processing plant and in which less than 10% of the animals are infected. These flocks usually don’t show a positive result during bacteriological monitoring. According to predictions of mathematical models this concerned 8% of the flocks in 2000. Totally 44% of the flocks were contaminated with Campylobacter. During the summer months the measured percentage of infected flocks may increase to over 60%. The measured percentage of infected flocks slowly decreased over the years from 36% in 2000 to 29% in 2003 [16].

• In an epidemiological study [2] it was shown that contamination of broiler flocks increased with the age of the animals, with the number of houses on a farm, the presence of other animals on the farm or direct surroundings and the access of children to houses when no specific clothing was worn. There was strong seasonal variation with peak prevalence in summer. The causes for this seasonal variation have not yet been elucidated.

• Modeling of data from infection experiments of experimental broiler flocks showed that only after more than a week after introduction of contamination the prevalence increases to such a level that this is detectable. After about two weeks nearly all animals are infected. The animals stay infected until the processing age

(approximately 42 days).

• Modeling of data from intensive monitoring of ten broiler farms showed that the more houses there are on the farm, the larger the probability of infection with

Campylobacter is. The infection of a flock is also higher when the previous flock was infected as well. This statistical association does not necessarily point to survival of Campylobacters in the broiler house between two cycles. There can also be other causes such as the overall level of farm hygiene and pressure of infection.

(19)

There are possibilities to reduce infection on the farm. In the CARMA project the following possibilities have been studied: discontinue thinning, discontinue keeping several species of farm animals on one farm (so-called mono-species farms),

improved hygiene and phage therapy, see Figure 4 and Table 3. The first two possibilities do not contribute – with constant production – to reducing the risk for the consumer. Improved hygiene is potentially a very effective measure, but it is as yet unclear which measures are achievable and effective in practice. Therefore is it not easy to assess the costs. Phage therapy is an experimental method which could lead to a significant reduction of risk for the consumer at relatively low costs.

0% 20% 40% 60% 80% 100% phage therapy farm hygiene Risk reduction no test culture dipstick PCR

Figure 4. Risk reduction for the consumer through interventions on the farm.

The bars show the predicted reduction of the number of cases of illness as a result of intervention as percentage of the incidence in the baseline model. This reduction concerns Dutch consumers of broiler meat originating from the Dutch production chain. The error bars show the results of sensitivity analyses. These sensitivity analyses are only performed for the interventions without scheduled treatment (no test). The effects of discontinuing thinning and of mono-species farms are negligible and therefore not portrayed.

To calculate the costs of interventions on the farm a base model of the economic situation in the broiler industry in the Netherlands has been created.

• In the year 2000 there were about 1,100 broiler farms in the Netherlands with an average occupation of 58,000 animals. Totally 420 million animals (840 million kilo live weight) were produced. On average there were 2.2 houses per farm. A round lasts about six weeks, followed by a week of vacancy, so there are about seven rounds per year. Because of the avian influenza crisis and the bad economical situation

(international competition, manure legislation) the poultry stock and the number of poultry farms have decreased in the last few years.

• The estimated average labor income of a broiler farm over the years 1993-2003 was 11,400 euro per farm per year. The average income of labor of all Dutch broiler farms was about 11.5 million euro per year. There are significant differences between the years, of a negative labor income of 31 million euro to a positive labor income of 47 million euro. These differences are connected to fluctuations in the farm-to-gate price that the farmer receives upon delivery to the processing plant.

(20)

The THINNING of a broiler flock is associated with a clearly larger probability of

contamination of the remaining animals with Campylobacter. There is a significantly higher percentage contaminated slaughter flocks after thinning. But it is unclear if this effect can be explained adequately from the higher age of the animals or if it is a matter of additional risk as a consequence of thinning. Even if during thinning a contamination would always be introduced, the risk for the consumer is negligible according to model predictions because in the relatively short time interval between partial and full depopulation only a small proportion of animals will be infected.

• Poultry farmers in the Netherlands use an all-in all-out system. This means that all houses are populated at the same moment by day-old chicks while all birds are also simultaneously transported to the processing plant. This transport however can be spread over several points of time, so that one house flock leads to several processing flocks. There are different reasons for this practice. First of all, the market asks for animals of different slaughter weights. Secondly, the density of broilers in a house may become too large.

• In the base year 2000 in about 40% of the cases a house flock was being transported to the processing plant as a whole. In the other cases this process took place in two or more phases. During thinning an average of 22% of the house flock was transported to the processing plant, about one week after followed by depopulation of the remaining animals.

• During thinning many people and materials access a house. Because of this there is an increased risk of contamination of the house with Campylobacter

(Molecular)-microbiological research has shown [9, Viv Allen, University of Bristol, UK; pers. comm.] that after thinning, more often a contamination with new Campylobacter types can be found, that are sometimes also found on material that was brought into the house such as crates and boots.

• In epidemiological studies contradictory results are found in relation to the risk of thinning. Usually a significantly increased risk is found, but sometimes this increased risk can be totally attributed to the aging of the animals, and there is no independent contribution of thinning.

• According to the mathematical transmission model it takes about two weeks before infection of a few broilers in a flock has spread to the majority of the animals. Even if it is assumed that during thinning a contamination in the house is always introduced (the most unfavorable scenario), than after a week only a small percentage of the animals (approximately 1%) from these houses is infected with Campylobacter. The associated risk for the consumer is negligible. Furthermore, if production remains constant, the number of broiler houses will have to increase. If these are built on the same premises, then the risk of infection per flock also increases.

• Given the – even in the most unfavorable scenario– minor contribution of thinning to the health risk of the consumer, to discontinue thinning is not an efficient measure.

Having SEVERAL TYPES OF FARM ANIMALS on one farm appears in epidemiological

studies to be associated with an increased risk of Campylobacter infection of broilers. However the farm model proves that discontinuing mixed farms and changing to monospecies farms does not, at a constant production, contribute to the reduction of the prevalence of contaminated flocks.

• With several types of farm animals on one farm the risk of Campylobacter

contamination of the broilers is 1.8 times higher than on farms with only broilers [3]. Mixed farms usually have 1 or 2 houses with broilers.

(21)

• In a hypothetical cross-over scenario farms with other farm animals would stop keeping 1 or 2 (on average 1.6) houses with broilers and specialized broiler farms would take over this production.

• As a result of this cross-over the production of broiler chickens remains about the same. The number of farms with broilers would decrease from approximately 1,100 to approximately 800, but the average size of the farms involved increases from 1.6 to 3.6 houses per farm.

• The risk of a flock being infected with Campylobacter increases when there are more houses on a farm. Because of the increase of the average farm size the cross-over scenario leads not to a decrease of the expected prevalence of Campylobacter contaminated flocks, but to a slight increase instead.

Further improving the HYGIENE on a farm is seen as an important measure to reduce

Campylobacter infection in broiler flocks. However it is still unclear in which way a broiler flock becomes infected. So it is not yet clear which measures need to be taken exactly and which effects these measures will have. Hygiene measures can be very effective if they result in a considerable reduction of the influx of Campylobacter. Which percentage is feasible in practice, is unknown. The costs are difficult to estimate but can vary between 8 and 63 million euro per year.

• The effect of hygiene measures on the influx of Campylobacter in a house is unclear. Therefore a sensitivity analyses is chosen. In theory it is calculated what the effect of an intervention could be, though in practice this is still unknown. A decrease of the probability of influx is assumed (both between rounds as between houses) with 5, 10 and 25%. It is striking that a reduction of the influx with 25% seems sufficient to almost completely resolve the infection because of non-linear feedback mechanisms. A smaller reduction of the probability of contamination with 5 or 10% reduces the calculated

prevalence of highly contaminated flocks from approximately 25% to approximately 19 respectively 14%.

• As a result of the predicted lower prevalence of contaminated flocks the risk for the consumer of chicken meat produced in the Netherlands decreases with 22-94% (if the influx decreases with 5-25%). Taking the import in

consideration it results in a reduction of 2,300-10,000 cases of illness in the Netherlands.

• The costs of extra hygiene measures are difficult to estimate. For the control of

Salmonella Java (which 2-3% of all farms have to deal with) a protocol has

been developed of which it was expected that some elements would also have an effect on Campylobacter. It concerns mainly the sealing of cracks and seams and extra cleaning and disinfecting between rounds. The costs of these measures can vary between 8 to upwards of 60 million euro per year,

including the opportunity costs of extra labor efforts of the farmers. The costs mainly depend on the frequency in which the measures need to be repeated. The costs are also lower when the measures only need to be applied on farms with an infected flock.

• The cost-utility of hygiene measures strongly depend on the assumptions made. In the most favorable case (reduction influx 25% and costs 8 million euro per year) the cost-utility ratio is 48,000 euro per averted DALY. With less favorable assumptions the cost-utility ratio is much more unfavorable.

(22)

PHAGE THERAPY is an experimental method by which the animals are treated with a

specific bacteriophage two days prior to processing. Bacteriophages are bacterial viruses that selectively inactivate Campylobacter but that are not dangerous to human health. If the effects of small-scale experiments would be confirmed in practice, than phage therapy would be a relatively inexpensive method to reduce risks for the

consumer. The efficiency is best when phage therapy is combined with a sensitive method to detect infected flocks. Such a method is presently being developed, but not yet available for use in practice.

• In experiments with phage therapy it was found that the concentration of

Campylobacter in the feces was about 2 log-units (100 times) lower at the moment of processing than in the feces of untreated animals. It is not known if the external contamination of the animals changes because of the phage therapy.

• Assuming a reduction of the concentration in the feces with 2 log-units would, according to model predictions, reduce the risk for the consumer of Dutch meat with approximately 75%. Taking import into account this leads to approximately 7,000 prevented cases of gastro-enteritis in the Netherlands.

• Assuming that the reduction of the concentration in the feces is only 1 log-unit, than the risk reduction is limited to approximately 45%. Assuming however that there is also reduction of the contamination of the exterior of chickens with 1 log-unit, than the risk reduction for consumers could be approximately 90% (see Figure 4). • The costs of phage therapy are not known, because there are no products on the

market yet. Judging from the price of existing treatments with antimicrobials they are estimated at a value between 0.0025 and 0.035 euro per treated animal with a most likely value of 0.02 euro. If all flocks that are to be processed are treated, than the estimated costs would be between 1 and 13 million euro per year with a most likely value upwards of 7 million euro per year.

• It is also possible to only treat the flocks infected with Campylobacter. A test of the Campylobacter status of the processing flocks needs to be performed. The costs of this test are estimated at 0.6 - 1 million euro per year, but the costs of phage therapy decrease with a factor of three. The total costs than decrease to upwards of 4.0 million euro per year, which is more than half cheaper than when all flocks are being treated. However, because not all infected flocks are recognized as such, the effectiveness decreases. Depending on the chosen test method according to the model predictions the risk for the consumer decreases with about 50 - 70%, see Figure 4.

• The costs of phage therapy are estimated at approximately 60,000 euro per averted DALY, when all animals are treated. With treatment of only positively tested animals the costs are 35,000 - 50,000 euro per averted DALY, with the assumption that the costs are 0.02 euro per treated broiler. These costs are however uncertain.

• It is yet unclear if long-term use of phage therapy will lead to selection of resistant campylobacters. Also the effectiveness in practice has not been proven yet.

During TRANSPORT of the animals from the farm to the processing plant

contamination can occur as a result of contaminated crates, trucks and the like. However the time between contamination and processing is very short so that only a few animals can be infected and then also with relatively low numbers. The risk for the consumer is therefore negligible. The transport phase is therefore not further included in the calculations.

(23)

6. Interventions at the processing plant

Because contamination of broilers with Campylobacter in the farm phase can not be completely prevented additional measures can be considered to reduce contamination of chicken meat. Measures regarding limiting cross-contamination during processing can be considered, as well as the separate processing of infected and non infected animals and the reduction of the level of contamination of meat. Of these possibilities reduction of fecal leakage during processing and a combination of scheduled

treatment and decontamination of carcasses with chemicals appear to be the most cost-effective options, see Figure 5 and Table 3. A reliable and fast test protocol needs to be available to detect infected flocks and the decontamination needs to be

technologically optimized.

• Animals from an infected flock are usually heavily contaminated with

Campylobacter. Almost all animals carry the infection in the gastrointestinal tract. An exception is the animals from a flock that were contaminated shortly before

processing. In such flocks, according to the farm model, the prevalence of

contaminated animals can be (much) lower than 10%. Such flocks are not recognized as contaminated during the usual testing procedures.

• Tests upon delivery at the processing plant show in contaminated flocks a median Campylobacter concentration of 106 colony forming units (cfu) per gram feces. All animals also appear to be contaminated externally (median 8 x 106 cfu per animal). • To quantify the spreading of Campylobacter during processing a mathematical model

has been formulated. This model describes the dynamics of the Campylobacter contamination in a “typical” processing plant in the base year 2000. Interventions in the processing phase can be simulated by changing one or more parameters in the model. The parameter values of the base model are based on structured interviews with experts because suitable measurement data were hardly available.

• According to the processing model the external contamination of a carcass from a contaminated flock declines during the whole process from 8 x 106 cfu per carcass to 3 x 104 cfu per carcass (median values). The largest decline takes place during scalding.

• A model was also created for the transmission of contamination from the exterior of a carcass to the fillets of chicken breasts during cutting. This model predicts that

approximately 100 cfu (median) will be present on a contaminated fillet. There is large variation of the contamination of individual fillets, between 0 and 104 cfu per fillet. The predicted contamination is lower than was found in practice measurements during the summer of 2004, with a median of 2000 cfu per fillet.

• The final contamination of a fillet of chicken breast depends to an approximately equal degree on the prevalence of contaminated animals within a flock, the

concentration of Campylobacter in the feces and the numbers on the exterior of an animal.

(24)

0% 20% 40% 60% 80% 100% prepared meat

product freezing irradiation crust freezing decont. dip and spray decont. dip decont. scald tank red. fecal leakage

Risk reduction

no test culture dipstick PCR

Figure 5. Risk reduction for the consumer by interventions at the processing plant.

The bars show the predicted reduction of the number of cases of illness as a result of intervention as percentage of the incidence in the baseline model. This reduction concerns Dutch consumers of broiler meat originating from the Dutch production chain. The error bars show the results of sensitivity analyses. These sensitivity analyses are only performed for the interventions without scheduled treatment (no test).

Newly developed and patented equipment aims to force a small amount of feces out of the cloaca by means of pressure on the abdomen, and to then spray the feces away with a pulse of water. The aim is to limit FECAL LEAKAGE during scalding and

defeathering. Calculations show that this method can be very cost-effective.

• It is not exactly known what the effect of this method is on the fecal leakage from the carcass in the next processing steps. For the model predictions, it appears not to make a substantial difference what is being assumed. The processing model predicts that the number of campylobacters on a chicken carcass after cooling will decline with a factor 3 to 10.

• Preliminary measurements in practice (by order of a manufacturer) support the model predictions.

• Reduction of fecal leakage would, according to model predictions lead to a maximum of 80% reduction of the risk for the consumer of Dutch meat. This equals to

approximately 9,000 prevented cases of gastro-enteritis per year, taking import into account.

(25)

• The annual costs of this intervention are approximately 1 million euro per year. In case the maximum risk reduction is reached the cost-utility ratio will be -15,000 euro per averted DALY. In other words, considered from the perspective of the whole Dutch society this intervention brings in net money. There is inevitably a difference between those who have to make the costs and those who profit from the benefits, see also Chapter 9.

• There is still very little practical information about the effectiveness of this technique and there are doubts whether the method will be successful under all circumstances. Further research in practice is therefore necessary to evaluate the model predictions.

DECONTAMINATION OF THE WATER IN THE SCALDING TANK is aimed at preventing

cross-contamination between carcasses during the scalding process. This measure provides a small contribution to the reduction of the risk and is relatively expensive.

• A broiler from a contaminated flock is externally contaminated with (median) 8 x 106_{cfu Campylobacter. As a result of the scalding process this number decreases} on average with a factor 20.

• In case a decontamination agent is added to the scald water (for example lactic acid in a concentration of 2.5%), then according to the literature, the external contamination of a carcass after scalding decreases with a factor up to 100.

• However, with defeathering the external contamination of a carcass increases. In comparison to this increase, the decrease of the external contamination in the scalding tank is negligible. This intervention therefore has only little effect on the risk for the consumer. The costs (approximately 13 million euro per year) are relatively high. So there is an unfavorable cost-utility ratio (810,000 euro per averted DALY).

Tracking contaminated animals before processing and taking additional measures with regard to contaminated flocks can contribute to reducing the contamination of chicken meat with Campylobacter. With LOGISTIC PROCESSING positively tested flocks

are processed at the end of the day or at a different location with an aim to prevent cross-contamination between infected and non-infected flocks. In case logistic

processing is introduced as the only measure the calculated risk for the consumer will not change substantially.

• The processing model predicts that at most a few carcasses from an uninfected flock become positive because of cross-contamination. Also the numbers of campylobacters on these carcasses will be considerably lower than on carcasses of flocks already infected upon delivery at the processing plant.

• Because the possibility of human infection is directly associated with the number of campylobacters to which the consumer is exposed, the low numbers on cross-contaminated carcasses imply that the prevention of cross-contamination by logistic processing contributes little to reduction of the public health risk.

• In Iceland it is obligatory to freeze meat when it is known that the flock was contaminated with Campylobacter1, based on testing of feces at the farm. Also for verification, coecum samples before processing and neck-skin samples after

processing are tested. Sometimes a flock is found of which the coecum samples are negative but the neck-skin samples are positive. These flocks are probably cross-contaminated at the processing plant. Measurements of those flocks confirm the model predictions of low numbers of contaminated carcasses, on which only low numbers of campylobacters are found (1-2 log cfu/carcass). The Icelandic authorities have decided that additional measures for these flocks are not necessary [Jarle Reiersen, Chief Veterinary Office of Iceland, pers. comm.].

(26)

With SCHEDULED TREATMENT positively tested flocks are not used for production of

fresh meat. The products can be diverted to the meat-processing industry, where they can undergo a heat treatment or otherwise to reduce the contamination. The

effectiveness of this measure strongly depends on the accuracy with which the

contaminated flocks can be detected. The protocol based on the culture method which is currently used to detect contamination of broilers with Campylobacter is not

sensitive enough to make scheduled treatment successful.

• The probability to actually detect an infected flock depends on the time between the infection of the flock (the first chicken) and the moment of sampling, the amount of samples and the sensitivity of the laboratory method.

• When a flock was contaminated two weeks or more before sampling, more than 50% of the broilers are infected. In that case the usual number of 10 samples is sufficient to detect the infection. When a flock was contaminated shorter before sampling, less than 10% of the broilers are infected. In that case the probability of detection is smaller than 5% and even with a random check of 100 samples the probability of detection is even smaller than 40%.

• The culture method can only be carried out in specialized laboratories, to where the samples need to be transported. Campylobacters are very sensitive bacteria that are easily damaged during transport. Based on experiences abroad (Norway) we estimate that only 50-75% of all contaminated samples are recognized as such, when sampling takes place a week prior to processing.

• When using the culture method, the time needed for transport of samples (with regular mail), for the analysis and for planning the processing is, at least, one week. During this week flocks can still get infected, but the significance of this seems minor. After all, during a timeframe of a few days the percentage of contaminated chickens in a flock will still be low. It is however possible that a flock is already contaminated during the moment of sampling, but not detectable because the percentage of infected chickens is still too small. A week later however all chickens among this flock are infected.

• If products from positively tested flocks using the culture method could be made totally Campylobacter free, the risk for the consumer would decrease with 67% (uncertainty margin 46-82%), see Figure 6.

• The sensitivity of a test protocol based on the culture method can probably be improved via courier transport and if necessary refrigerated. To what extent this is really the case is unknown. We estimate that the sensitivity can improve to 90%. • Even with this sensitivity still 1% of the contaminated flocks will not be recognized as

such and end up in the fresh meat channel. Scheduled treatment with this detection method reduces the risk for the consumer of meat produced in the Netherlands with a maximum of 87%.

(27)

risk reduction scheduled treatment with elimination of positive flocks 0% 20% 40% 60% 80% 100% PCR dipstick culture

Figure 6. Theoretical risk reduction (best estimate and uncertainty range) of scheduled treatment with three different test protocols.

The figure shows the maximum achievable risk reduction for the consumer with the use of scheduled treatment based on analysis of samples on the farm using a culture method, a dipstick method and a PCR method. In each case 10 sample are obtained per house which are pooled into one test sample. The bars show the most likely estimate; the error bars the results of the sensitivity analyses.

Alternatives for the culture method are being developed, but are not yet validated nor commonly available. They could be available in one or two years and when their sensitivity lives up to the expectations they can be a good base for scheduled treatment.

• In the Netherlands a dipstick method is being developed that can be used by the farmer himself to establish the Campylobacter status of the living flocks. Because transport is not necessary, and the result is obtained immediately, the test can be performed only two days before processing. This reduces the probability that a flock gets heavily infected in the time span between sampling and processing. It has been estimated that this method will recognize 70-95% of all contaminated samples as such. There is however no practical experience yet. In case the results are confirmed in practice, scheduled treatment based on the results of this test protocol could reduce the risk for the consumer with a maximum of 70-90%, see Figure 6.

• In Denmark a molecular (PCR) method is in use by a number of companies. This test must be performed by a specialized laboratory. It is assumed that samples are taken five days before processing. Practical information is not yet available. Based on the literature [11] it is expected that 90-100% of the contaminated flocks can be detected. The risk for the consumer then decreases with a maximum of 84-93%, see Figure 6. The risk will never be totally eliminated because there will always be lightly

contaminated flocks that will escape the test.

SCHEDULED TREATMENT to meat processing can result in a significant decrease of

income for the chicken meat industries because fresh meat sells at higher prices. There is also a risk of permanent market loss. With the current prevalence there is not enough supply of negatively tested flocks during the summer to satisfy the demand for fresh meat.

• Scheduled treatment can make it more difficult for the industry to deliver the desired products on time. Also extra transport costs and less efficient use of processing lines can result. It is unclear to what extent these effects will show in reality, but the potential economical damage may be substantial.

• During the summer 60% of the flocks are tested positive using the culture method. The actual prevalence is much higher. Of the total production of chicken meat in the Netherlands approximately 2/3 is fresh meat. Thus there is an insufficient supply of negatively tested flocks during the summer.